OOD Detection on CIFAR-10 In-Distribution (vs SVHN, CelebA, C100, Tex)
0.951Performance vs SVHNUFCOD
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| UFCODTraining Data Regime=Few-shot2026.05 | 0.951 | 0.965 | 0.547 | 0.883 | |
| DiffPathTraining Data Regime=Full-data2026.05 | 0.91 | 0.897 | 0.59 | 0.923 | |
| VAEBMTraining Data Regime=Full-data2026.05 | 0.695 | 0.26 | 0.48 | 0.685 | |
| MSMATraining Data Regime=Full-data2026.05 | 0.611 | 0.535 | 0.51 | 0.518 | |
| IGEBMTraining Data Regime=Full-data2026.05 | 0.581 | 0.539 | 0.503 | 0.448 | |
| DDPM-OODTraining Data Regime=Full-data2026.05 | 0.085 | 0.539 | 0.52 | 0.396 | |
| NLLTraining Data Regime=Full-data2026.05 | 0.046 | 0.559 | 0.555 | 0.579 | |
| LMDTraining Data Regime=Full-data2026.05 | 0.046 | 0.524 | 0.529 | 0.586 | |
| DoSTraining Data Regime=Full-data2026.05 | 0.04 | 0.582 | 0.566 | 0.571 |